Re: [igraph] Query regarding modularity calculations

[Gábor Csárdi]

Dear Vijaykumar Yogesh Muley,

On Fri, Jan 15, 2010 at 4:58 PM, Vijaykumar Muley <address@hidden> wrote:
> Dear all,
>
> I tried to run modularity calculations for a graph which has 4000 nodes and
> 60,000 interactions. I performed these calculations using modularity
> function of the igraph packages the R code is as follow
>
> x is a graph
>
> b <- edge.betweenness.community( x )
> comm <- community.to.membership( x, b$merges, steps=10)
> modularity=modularity(x, comm$membership)
>
> but I am getting negative values for modularity moreover it took around 5
> days.

as I already wrote you, it is not the modularity calculation that
takes a long time, but the edge-betweenness based community structure
detection. The modularity calculation should be fast.

Moreover, the negative value is normal, modularity is not necessarily
positive. If you just take 10 merges in such a big graph, as you did,
then you will have very poor communities and the modularity score will
be negative.

Best Regards,
Gabor

> am I doing anything wrong ? can you suggests me where I am going wrong and
> the estimated time to perform these calculations ?
> is there alternate ways to calculate modularity ?
> Please, do help me to  overcome this problem.
>
>
> Thanks and Regards
>
>
> --
> Vijaykumar Yogesh Muley
>
> Senior  Research Fellow,
> Dr. Akash Ranjan's Group,
> Computational and Functional Genomics Laboratory (CFG),
> Centre for DNA Fingerprinting and Diagnostics (CDFD),
> Bldg.7, Gruhakalpa
> 5-4-399/B, Nampally
> Hyderabad- 500001
> India
>
> Ph.No: +91 9441189013
> E-mail : address@hidden
>
> Life : A journey from Cell to Cell
> ----------------------------------------------
>
> _______________________________________________
> igraph-help mailing list
> address@hidden
> http://lists.nongnu.org/mailman/listinfo/igraph-help
>
>



-- 
Gabor Csardi <address@hidden>     UNIL DGM